Primary Convener:
Simon Driscoll, University of Cambridge

Convener:
Sara Shamekh, New York University
Akshay Subramaniam, NVIDIA Corporation
Aakash Sane, Princeton University

Early Career Convener:
Karan Jakhar, Rice University

Chair:
Sara Shamekh, New York University
Simon Driscoll, University of Reading
Aakash Sane, Brown University
Karan Jakhar, Rice University
Akshay Subramaniam, NVIDIA Corporation

Machine learning is reshaping the representation of complex physical processes in Earth system models, offering new avenues for parameterization, emulation, and hybrid modeling. This session focuses on the use of machine learning to emulate computationally expensive or unresolved processes, accelerate physical simulations, and improve representation across domains such as convection, turbulence, radiation, hydrology, sea ice, and other components of the Earth system. Topics include (but are not limited to): – Subgrid-scale parameterization via machine learning – Emulators of physical processes, model components, or whole models – Hybrid ML-physics modeling frameworks – Physics-informed neural networks, neural operators, and differentiable programming – Reinforcement learning – Cross-domain applications (atmosphere, ocean, cryosphere, land). By bringing together approaches that integrate data-driven and physically-based methods, this session will provide a critical overview of current progress and emerging directions in the application of machine learning to parameterizations, emulation, and hybrid modeling across Earth system science.

Index Terms
0545 Modeling
1622 Earth system modeling
1942 Machine learning
4430 Complex systems

Cross-Listed:
IN - Informatics
C - Cryosphere
A - Atmospheric Sciences
OS - Ocean Sciences

Suggested Itineraries:
Machine Learning and AI

Neighborhoods:
1. Science Nexus